* Dari-Mattiacci & Fabbri - Respect for Property

clear
use Submission_Property_dataset.dta

count if treated==1
* 73
count if treated==0
*54

preserve
collapse (mean) DictatorTakeManna (sd) sdDictatorTakeManna=DictatorTakeManna if DictatorTakeManna!= ., by(treated)
generate n = 55 if treated==0
replace n = 73 if treated==1
generate hicoop = DictatorTakeManna + invttail(n-1,0.025)*(sdDictatorTakeManna / sqrt(n))
generate locoop = DictatorTakeManna - invttail(n-1,0.025)*(sdDictatorTakeManna / sqrt(n))
label define treated 0 "No Reform" 1 "Formalized Rights", replace
twoway (bar DictatorTakeManna treated if treated==1, barwidth(0.6) color(gs3)) (bar DictatorTakeManna treated if treated==0, barwidth(0.6) color(gs11)) (rcap hicoop locoop treated), ytitle() xtitle(, size(zero)) yscale(range(0 4)) xscale(range(-0.7 1.7)) ylabel(0(1)4, labels valuelabel ticks) xlabel(0(1)1, labels valuelabel ticks) ymtick(minmax) xmtick(minmax) title(Coins appropriated by the dictators) legend(off)
restore

sktest DictatorTakeManna
label define treated 0 "No Reform" 1 "Formalized Rights", replace
hist DictatorTakeManna if treated==0, bcolor(gs11)  xtitle(No Reform) disc freq normal
hist DictatorTakeManna if treated==1, bcolor(gs3) xtitle(Formalized Rights) disc freq normal 

ranksum DictatorTakeManna, by (treated)
ttest DictatorTakeManna, by (treated)
* the treatment significantly reduces stealing 

tab dtakemanna treated, column chi2
* sign 1% more zero-takers in treated

*****************
* CHECK RANDOMIZATION

ttest income10k, by(treated)
ranksum income10k, by(treated)

ttest education, by(treated)
ranksum education, by(treated)

tab male treated, column chi2

ttest age, by(treated)
ranksum age, by(treated)

ttest christian, by(treated)
ranksum christian, by(treated)

ttest otherrel, by(treated)
ranksum otherrel, by(treated)

ttest married, by(treated)

ttest monog, by(treated)

ttest yearsinvillage, by(treated)

preserve
collapse population distanceroad, by (treated session)
ttest population, by(treated)
ttest distanceroad, by(treated)
restore
* sign difference

***************
*** TABLE 1 ***
***************

* a) version with standard bootstrap 
est clear
reg DictatorTakeManna treated distanceroad population age male riskloss muslim christian, vce(bootstrap, cluster(session) reps(500) dots(1))
 est sto reg3

tobit DictatorTakeManna treated distanceroad population age male riskloss muslim christian , ll ul vce(bootstrap, cluster(session) reps(500) dots(1))
 est sto reg1 
 
reg DictatorTakeManna treated age education income10k finance male riskloss muslim christian distanceroad population, vce(bootstrap, cluster(session) reps(500) dots(1))
 est sto reg4
 
tobit DictatorTakeManna treated age education income10k finance male riskloss muslim christian distanceroad population, ll ul vce(bootstrap, cluster(session) reps(500) dots(1))
 est sto reg2
 
 estout reg3 reg1 reg4 reg2  using "../Elaborations/tables/table1-revision.tex", replace stardetach ///
cells("b(star fmt(3))" "se(par fmt(3))") substitute( _ "-" -cons "Constant") ///
indicate() nobase ///
style(tex) label collabels(,none) stats(N, labels("N.obs.")fmt(0)) ///
 prehead(\begin{tabular}{p{.45\textwidth}*{4}{r@{}l}} \hline \hline ///
"& \multicolumn{2}{c}{Model 1} & \multicolumn{2}{c}{Model 2}  \\") ///
posthead(\hline) prefoot(\hline) postfoot(\hline \multicolumn{9}{p{\textwidth}} ///
{\footnotesize{\textbf{Notes:} ///
Dependent variable: coins appropriated by the dictator. ///
Models 1 and 3: OLS regression, ///
models 2 and 4: left-censored Tobit regression. ///
All models include village fixed-effects. ///
Robust standard errors clustered at the village level. ///
Compared to models 1 and 2, models 3 and 4 controls for income, finance, and education levels. /// 
Controls include: age, ///
 gender, ///
religion, ///
estimation of risk preferences. /// 
Symbols $***$, $**$, and $*$ indicate significance at the 1\%, 5\% and 10\% level, respectively.}} \end{tabular}) ///
starlevels(* 0.10 ** 0.05 *** 0.01) ml(,none) eql(,none)


* b) Version with Wild cluster bootstrap

reg DictatorTakeManna treated distanceroad population age male riskloss muslim christian, vce(cluster session) 
boottest treated, svmat

tobit DictatorTakeManna treated distanceroad population age male riskloss muslim christian , ll ul vce(cluster session) 
boottest treated, svmat
 
reg DictatorTakeManna treated age education income10k finance male riskloss muslim christian distanceroad population, vce(cluster session) 
boottest treated, svmat
boottest education, svmat
boottest income10k, svmat
boottest finance, svmat
 
tobit DictatorTakeManna treated age education income10k finance male riskloss muslim christian distanceroad population, ll ul vce(cluster session)
boottest treated, svmat
boottest education, svmat
boottest income10k, svmat
boottest finance, svmat



*******************************************
**** TEST HYPOTHESIS EXPLAINING RESULTS ***
*******************************************


************
* 1) INCOME 
************

ranksum income10k, by(treated)
* no differences

ranksum education, by(treated)
* no differences

*****************************************************
*** TABLE 2   Income ***
*****************************************************

est clear

reg income10k treated  age male riskloss muslim christian distanceroad, vce(cluster session) 
boottest treated, svmat
est sto reg1

reg income10k treated  age education male riskloss muslim christian distanceroad, vce(cluster session)
boottest treated, svmat 
est sto reg11

 estout reg1 reg11  using "../Elaborations/tables/table2income.tex", replace stardetach ///
cells("b(star fmt(3))" "se(par fmt(3))") substitute( _ "-" -cons "Constant") ///
indicate() nobase ///
style(tex) label collabels(,none) stats(N, labels("N.obs.")fmt(0)) ///
 prehead(\begin{tabular}{p{.45\textwidth}*{2}{r@{}l}} \hline \hline ///
"& \multicolumn{2}{c}{Model 1} & \multicolumn{2}{c}{Model 2}  \\") ///
posthead(\hline) prefoot(\hline) postfoot(\hline \multicolumn{9}{p{\textwidth}} ///
{\footnotesize{\textbf{Notes:} ///
Dependent variable models 1 and 2: \emph{income} (thousands of XOF). ///
OLS regression with village fixed effects. ///
Robust standard errors clustered at the village level. ///
Compared to model 1, model 2 controls for education. /// 
Controls include age, ///
gender, ///
religion, ///
estimation of risk preferences. /// 
Symbols $***$, $**$, and $*$ indicate significance at the 1\%, 5\% and 10\% level, respectively.}} \end{tabular}) ///
starlevels(* 0.10 ** 0.05 *** 0.01) ml(,none) eql(,none)




************
* 2) EDUCATION
************

*****************************************************
*** TABLE 3   Education ***
*****************************************************

ranksum education, by(treated)
* no differences

est clear

reg education treated  age male riskloss muslim christian distanceroad, vce(cluster session) 
boottest treated, svmat
est sto reg2

reg education treated  age income10k male riskloss muslim christian distanceroad, vce(cluster session) 
boottest treated, svmat
est sto reg22

 estout reg2 reg22  using "../Elaborations/tables/table3education.tex", replace stardetach ///
cells("b(star fmt(3))" "se(par fmt(3))") substitute( _ "-" -cons "Constant") ///
indicate() nobase ///
style(tex) label collabels(,none) stats(N, labels("N.obs.")fmt(0)) ///
 prehead(\begin{tabular}{p{.45\textwidth}*{2}{r@{}l}} \hline \hline ///
"& \multicolumn{2}{c}{Model 1} & \multicolumn{2}{c}{Model 2}  \\") ///
posthead(\hline) prefoot(\hline) postfoot(\hline \multicolumn{9}{p{\textwidth}} ///
{\footnotesize{\textbf{Notes:} ///
Dependent variable models 1 and 2: \emph{education} (in years). ///
Left-censored Tobit regressions. ///
Robust standard errors clustered at the village level. ///
Compared to model 1, model 2 controls for income levels. /// 
Controls include age, ///
 gender, ///
religion, ///
estimation of risk preferences. /// 
Symbols $***$, $**$, and $*$ indicate significance at the 1\%, 5\% and 10\% level, respectively.}} \end{tabular}) ///
starlevels(* 0.10 ** 0.05 *** 0.01) ml(,none) eql(,none)




**************
* 3) FINANCE *
**************

tab finance treated, column chi2
* no difference in participation financial activities

******************************************
*** TABLE 4 access to financial markets***
******************************************

est clear

probit finance treated  age  male riskloss muslim christian distanceroad, vce(cluster session) 
boottest treated, svmat
est sto reg2

probit finance treated  age education income10k male riskloss muslim christian distanceroad, vce(cluster session) 
boottest treated, svmat

est sto reg22
 
 estout reg2 reg22  using "../Elaborations/tables/table4finance.tex", replace stardetach ///
cells("b(star fmt(3))" "se(par fmt(3))") substitute( _ "-" -cons "Constant") ///
indicate() nobase ///
style(tex) label collabels(,none) stats(N, labels("N.obs.")fmt(0)) ///
 prehead(\begin{tabular}{p{.45\textwidth}*{2}{r@{}l}} \hline \hline ///
"& \multicolumn{2}{c}{Model 1} & \multicolumn{2}{c}{Model 2}  \\") ///
posthead(\hline) prefoot(\hline) postfoot(\hline \multicolumn{9}{p{\textwidth}} ///
{\footnotesize{\textbf{Notes:} ///
Dependent variable models 1 and 2: \emph{finance}, dummy equal 1 if the respondent has requested a loan, a mortgage, or has participated in other financial activities in the previous 7 years. ///
Probit regressions with village level fixed effects. ///
Robust standard errors clustered at the village level. ///
Compared to model 1, model 2 controls for income and education. /// 
Controls include age, ///
 gender, ///
religion, ///
estimation of risk preferences. /// 
Symbols $***$, $**$, and $*$ indicate significance at the 1\%, 5\% and 10\% level, respectively.}} \end{tabular}) ///
starlevels(* 0.10 ** 0.05 *** 0.01) ml(,none) eql(,none)



**************
* 4) CONFLICTS
**************
 
 * Conflicts overall
tab conflict treated, column chi2

***************
*** TABLE 5 Conflicts ***
***************

est clear

probit conflict treated age male riskloss muslim christian distanceroad, vce(cluster session)
boottest treated, svmat
 est sto reg1

probit conflict treated age education income10k male riskloss muslim christian distanceroad, vce(cluster session)
boottest treated, svmat
 est sto reg2
 
 estout reg1 reg2  using "../Elaborations/tables/table5conflicts.tex", replace stardetach ///
cells("b(star fmt(3))" "se(par fmt(3))") substitute( _ "-" -cons "Constant") ///
indicate() nobase ///
style(tex) label collabels(,none) stats(N, labels("N.obs.")fmt(0)) ///
 prehead(\begin{tabular}{p{.45\textwidth}*{2}{r@{}l}} \hline \hline ///
"& \multicolumn{2}{c}{Model 1} & \multicolumn{2}{c}{Model 2}  \\") ///
posthead(\hline) prefoot(\hline) postfoot(\hline \multicolumn{9}{p{\textwidth}} ///
{\footnotesize{\textbf{Notes:} ///
Dependent variable models 1 and 2: \emph{conflict}, dummy equal 1 if the respondent has experienced a conflict in previous 7 years. ///
Probit regression with village fixed effects. ///
Robust standard errors clustered at the village level. ///
Compared to model 1, model 2 controls for income and education. /// 
Controls include age, ///
 gender, ///
religion, ///
estimation of risk preferences. /// 
Symbols $***$, $**$, and $*$ indicate significance at the 1\%, 5\% and 10\% level, respectively.}} \end{tabular}) ///
starlevels(* 0.10 ** 0.05 *** 0.01) ml(,none) eql(,none)


* Through which institution are conflicts addressed? - mostly no recourse to legal system
tab formalconflictresolution treated, column chi2
* no difference

* Are conflict solved? 
tab conflictsolved treated, column chi2
* no difference



***************************************************************************************
* 4) Altruism, beliefs regarding others' taking, social norms regarding redistribution*
***************************************************************************************

************
* Donation *
************

preserve
collapse (mean) Donation (sd) sdDictatorTakeManna=Donation if DictatorTakeManna!= ., by(treated)
generate n = 55 if treated==0
replace n = 73 if treated==1
generate hicoop = Donation + invttail(n-1,0.025)*(sdDictatorTakeManna / sqrt(n))
generate locoop = Donation - invttail(n-1,0.025)*(sdDictatorTakeManna / sqrt(n))
label define treated 0 "No Reform" 1 "Formalized Rights", replace
twoway (bar Donation treated if treated==1, barwidth(0.6) color(gs3)) (bar Donation treated if treated==0, barwidth(0.6) color(gs11)) (rcap hicoop locoop treated), ytitle() xtitle(, size(zero)) yscale(range(0 4)) xscale(range(-0.7 1.7)) ylabel(0(1)4, labels valuelabel ticks) xlabel(0(1)1, labels valuelabel ticks) ymtick(minmax) xmtick(minmax) title(Coins donated) legend(off)
restore

sktest Donation
label define treated 0 "No Reform" 1 "Formalized Rights", replace
hist Donation if treated==0, bcolor(g11) xtitle(No Reform) disc freq normal
hist Donation if treated==1, bcolor(g3) xtitle(Formalized Rights) disc freq normal 

ranksum Donation, by (treated)
ttest Donation, by (treated)
* difference not statistically significant

reg Donation treated age education income male riskloss muslim christian distanceroad, vce(cluster session)
boottest treated, svmat

tobit Donation treated age education income male riskloss muslim christian distanceroad, vce(cluster session)
boottest treated, svmat




*************************************************************************
* Beliefs of passive player regarding coind appropriated by the dictator*
*************************************************************************
preserve
use Property_beliefspassiveplayer.dta, clear
collapse (mean) PassivePlayerBeliefsManna (sd) sdDictatorTakeManna=PassivePlayerBeliefsManna, by(treated)
generate n = 55 if treated==0
replace n = 73 if treated==1
generate hicoop = PassivePlayerBeliefsManna + invttail(n-1,0.025)*(sdDictatorTakeManna / sqrt(n))
generate locoop = PassivePlayerBeliefsManna - invttail(n-1,0.025)*(sdDictatorTakeManna / sqrt(n))
label define treated 0 "No Reform" 1 "Formalized Rights", replace
twoway (bar PassivePlayerBeliefsManna treated if treated==1, barwidth(0.6) color(gs3)) (bar PassivePlayerBeliefsManna treated if treated==0, barwidth(0.6) color(gs11)) (rcap hicoop locoop treated), ytitle() xtitle(, size(zero)) yscale(range(0 4)) xscale(range(-0.7 1.7)) ylabel(0(1)4, labels valuelabel ticks) xlabel(0(1)1, labels valuelabel ticks) ymtick(minmax) xmtick(minmax) title(Beliefs regarding coins appropriated by the dictator) legend(off)
restore

***************************************************************
* Division of income -- Proportional vs. Egalitarian criterion*
***************************************************************

kwallis SN_egalitarian, by(treated)
kwallis SN_proportional, by(treated)


******************************************
* 5) Other beliefs supportive of markets *
******************************************

ttest Individualism, by (treated)
ranksum Individualism, by (treated)

tab Individualism treated, column chi2
***

***
ttest WorkEthics , by (treated)
ranksum WorkEthics , by (treated)

tab WorkEthics treated, column chi2
***

***
ttest Moneyimportance , by (treated)
ranksum Moneyimportance , by (treated)

***

* 1) The graph (histogram of 0-10) (Figure 4)

preserve
gen graph=DictatorTakeManna
collapse (sum) DictatorTakeManna, by(treated  graph)
gen Formalized_Rights=DictatorTakeManna if treated
gen No_Reform=DictatorTakeManna if treated==0
graph bar Formalized_Rights No_Reform, over(graph) asyvars bar(1, fcolor(gs3)) bar(2, fcolor(gs11)) legend(label(1 "Formalized Rights") ///
label (2 "No Reform") ) ytitle("Sum of coins taken")
restore


* 2) Negative binomial instead of Tobit (Table 3 appendix)

est clear

zinb DictatorTakeManna treated distanceroad population age male riskloss muslim christian, inflate(education age male riskloss muslim christian) vce(cluster session)
boottest treated, svmat

est sto reg3

zinb DictatorTakeManna treated distanceroad population age male riskloss muslim christian, inflate(population distanceroad education age male riskloss muslim christian) vce(cluster session)
est sto reg1
boottest treated, svmat


zinb DictatorTakeManna treated distanceroad population age male riskloss muslim christian income10k finance education, inflate(education age male riskloss muslim christian) vce(cluster session)
est sto reg4
boottest treated, svmat

zinb DictatorTakeManna treated distanceroad population age male riskloss muslim christian income10k finance education, inflate(distanceroad education age male riskloss muslim christian) vce(cluster session)
est sto reg2
boottest treated, svmat
boottest const, svmat

 
 estout reg3 reg1 reg4 reg2  using "../Elaborations/tables/table1bisNBR-revision.tex", replace stardetach ///
cells("b(star fmt(3))" "se(par fmt(3))") substitute( _ "-" -cons "Constant") ///
indicate() nobase ///
style(tex) label collabels(,none) stats(N, labels("N.obs.")fmt(0)) ///
 prehead(\begin{tabular}{p{.45\textwidth}*{4}{r@{}l}} \hline \hline ///
"& \multicolumn{2}{c}{Model 1} & \multicolumn{2}{c}{Model 2}  \\") ///
posthead(\hline) prefoot(\hline) postfoot(\hline \multicolumn{9}{p{\textwidth}} ///
{\footnotesize{\textbf{Notes:} ///
Dependent variable: coins appropriated by the dictator. ///
Models 1 and 3: OLS regression, ///
models 2 and 4: left-censored Tobit regression. ///
All models include village fixed-effects. ///
Robust standard errors clustered at the village level. ///
Compared to models 1 and 2, models 3 and 4 controls for income, finance, and education levels. /// 
Controls include: age, ///
 gender, ///
religion, ///
estimation of risk preferences. /// 
Symbols $***$, $**$, and $*$ indicate significance at the 1\%, 5\% and 10\% level, respectively.}} \end{tabular}) ///
starlevels(* 0.10 ** 0.05 *** 0.01) ml(,none) eql(,none)
 

* 4) Is the result more similar for villages that are closer to each other (or south/north)?    

*** ONLY SOUTH ***
preserve
keep if south
est clear

reg DictatorTakeManna treated distanceroad population age male riskloss muslim christian, vce(cluster session) 
boottest treated, svmat
 est sto reg3

tobit DictatorTakeManna treated distanceroad population age male riskloss muslim christian , ll ul vce(cluster session) 
boottest treated, svmat

 est sto reg1 
 
reg DictatorTakeManna treated age education income10k finance male riskloss muslim christian distanceroad population, vce(cluster session) 
boottest treated, svmat

 est sto reg4
 
tobit DictatorTakeManna treated age education income10k finance male riskloss muslim christian distanceroad population, ll ul vce(cluster session) 
boottest treated, svmat

 est sto reg2
 
 estout reg3 reg1 reg4 reg2  using "../Elaborations/tables/tablesouthonly-revision.tex", replace stardetach ///
cells("b(star fmt(3))" "se(par fmt(3))") substitute( _ "-" -cons "Constant") ///
indicate() nobase ///
style(tex) label collabels(,none) stats(N, labels("N.obs.")fmt(0)) ///
 prehead(\begin{tabular}{p{.45\textwidth}*{4}{r@{}l}} \hline \hline ///
"& \multicolumn{2}{c}{Model 1} & \multicolumn{2}{c}{Model 2}  \\") ///
posthead(\hline) prefoot(\hline) postfoot(\hline \multicolumn{9}{p{\textwidth}} ///
{\footnotesize{\textbf{Notes:} ///
Dependent variable: coins appropriated by the dictator. ///
Models 1 and 3: OLS regression, ///
models 2 and 4: left-censored Tobit regression. ///
All models include village fixed-effects. ///
Robust standard errors clustered at the village level. ///
Compared to models 1 and 2, models 3 and 4 controls for income, finance, and education levels. /// 
Controls include: age, ///
 gender, ///
religion, ///
estimation of risk preferences. /// 
Symbols $***$, $**$, and $*$ indicate significance at the 1\%, 5\% and 10\% level, respectively.}} \end{tabular}) ///
starlevels(* 0.10 ** 0.05 *** 0.01) ml(,none) eql(,none)
restore

preserve
keep if south
collapse (mean) DictatorTakeManna (sd) sdDictatorTakeManna=DictatorTakeManna if DictatorTakeManna!= ., by(treated)
generate n = 55 if treated==0
replace n = 73 if treated==1
generate hicoop = DictatorTakeManna + invttail(n-1,0.025)*(sdDictatorTakeManna / sqrt(n))
generate locoop = DictatorTakeManna - invttail(n-1,0.025)*(sdDictatorTakeManna / sqrt(n))
label define treated 0 "No Reform" 1 "Formalized Rights", replace
twoway (bar DictatorTakeManna treated if treated==1, barwidth(0.6) color(gs3)) (bar DictatorTakeManna treated if treated==0, barwidth(0.6) color(gs11)) (rcap hicoop locoop treated), ytitle() xtitle(, size(zero)) yscale(range(0 4)) xscale(range(-0.7 1.7)) ylabel(0(1)4, labels valuelabel ticks) xlabel(0(1)1, labels valuelabel ticks) ymtick(minmax) xmtick(minmax) title(Southern Villages - Coins appropriated) legend(off)
restore

preserve
keep if south
sktest DictatorTakeManna
label define treated 0 "No Reform" 1 "Formalized Rights", replace
hist DictatorTakeManna if treated==0, bcolor(gs11) xtitle(No Reform) xscale(range(0 10)) xlabel(0(2)10) disc freq normal
restore

preserve
keep if south
sktest DictatorTakeManna
label define treated 0 "No Reform" 1 "Formalized Rights", replace
hist DictatorTakeManna if treated==1, bcolor(gs3) xtitle(Formalized Rights)  xscale(range(0 10)) xlabel(0(2)10) disc freq normal 
restore

preserve
keep if south
ranksum DictatorTakeManna, by (treated)
ttest DictatorTakeManna, by (treated)
* the treatment significantly reduces stealing 
tab dtakemanna treated, column chi2
* sign 1% more zero-takers in treated
restore

******************
*** ONLY NORTH ***
******************

preserve
keep if south==0
est clear

reg DictatorTakeManna treated distanceroad population age male riskloss muslim christian, vce(cluster session) 
boottest treated, svmat
 est sto reg3

tobit DictatorTakeManna treated distanceroad population age male riskloss muslim christian , ll ul vce(cluster session) 
boottest treated, svmat
 est sto reg1 
 
reg DictatorTakeManna treated age education income10k finance male riskloss muslim christian distanceroad population, vce(cluster session) 
boottest treated, svmat
 est sto reg4
 
tobit DictatorTakeManna treated age education income10k finance male riskloss muslim christian distanceroad population, ll ul vce(cluster session) 
boottest treated, svmat
 est sto reg2
 
 estout reg3 reg1 reg4 reg2  using "../Elaborations/tables/tablenorthonly-revision.tex", replace stardetach ///
cells("b(star fmt(3))" "se(par fmt(3))") substitute( _ "-" -cons "Constant") ///
indicate() nobase ///
style(tex) label collabels(,none) stats(N, labels("N.obs.")fmt(0)) ///
 prehead(\begin{tabular}{p{.45\textwidth}*{4}{r@{}l}} \hline \hline ///
"& \multicolumn{2}{c}{Model 1} & \multicolumn{2}{c}{Model 2}  \\") ///
posthead(\hline) prefoot(\hline) postfoot(\hline \multicolumn{9}{p{\textwidth}} ///
{\footnotesize{\textbf{Notes:} ///
Dependent variable: coins appropriated by the dictator. ///
Models 1 and 3: OLS regression, ///
models 2 and 4: left-censored Tobit regression. ///
All models include village fixed-effects. ///
Robust standard errors clustered at the village level. ///
Compared to models 1 and 2, models 3 and 4 controls for income, finance, and education levels. /// 
Controls include: age, ///
 gender, ///
religion, ///
estimation of risk preferences. /// 
Symbols $***$, $**$, and $*$ indicate significance at the 1\%, 5\% and 10\% level, respectively.}} \end{tabular}) ///
starlevels(* 0.10 ** 0.05 *** 0.01) ml(,none) eql(,none)

restore


preserve
keep if south==0
collapse (mean) DictatorTakeManna (sd) sdDictatorTakeManna=DictatorTakeManna if DictatorTakeManna!= ., by(treated)
generate n = 55 if treated==0
replace n = 73 if treated==1
generate hicoop = DictatorTakeManna + invttail(n-1,0.025)*(sdDictatorTakeManna / sqrt(n))
generate locoop = DictatorTakeManna - invttail(n-1,0.025)*(sdDictatorTakeManna / sqrt(n))
label define treated 0 "No Reform" 1 "Formalized Rights", replace
twoway (bar DictatorTakeManna treated if treated==1, barwidth(0.6) color(gs3)) (bar DictatorTakeManna treated if treated==0, barwidth(0.6) color(gs11)) (rcap hicoop locoop treated), ytitle() xtitle(, size(zero)) yscale(range(0 4)) xscale(range(-0.7 1.7)) ylabel(0(1)4, labels valuelabel ticks) xlabel(0(1)1, labels valuelabel ticks) ymtick(minmax) xmtick(minmax) title(Northern Villages - Coins appropriated) legend(off)
restore

preserve
keep if south==0
sktest DictatorTakeManna
label define treated 0 "No Reform" 1 "Formalized Rights", replace
hist DictatorTakeManna if treated==0, bcolor(gs11) xtitle(No Reform)  xscale(range(0 10)) xlabel(0(2)10)   disc freq normal 
restore

preserve
keep if south==0
sktest DictatorTakeManna
label define treated 0 "No Reform" 1 "Formalized Rights", replace
hist DictatorTakeManna if treated==1, bcolor(gs3) xtitle(Formalized Rights)  disc freq normal 
restore


preserve 
keep if south==0
ranksum DictatorTakeManna, by (treated)
ttest DictatorTakeManna, by (treated)
* the treatment significantly reduces stealing 
tab dtakemanna treated, column chi2
* sign 1% more zero-takers in treated
restore


********************************************


* 1) Donation game: replicate the estimates like for the taking game (version with standard bootstrap and version with Wild cluster bootstrap)

est clear

reg Donation treated distanceroad population age male riskloss muslim christian, vce(bootstrap, cluster(session) reps(500) dots(1))
 est sto reg3

reg Donation treated distanceroad population age male riskloss muslim christian, vce(cluster session)
boottest treated, svmat 

tobit Donation treated distanceroad population age male riskloss muslim christian , ll ul vce(bootstrap, cluster(session) reps(500) dots(1))
estat bootstrap, all
 est sto reg1 
 
 tobit Donation treated distanceroad population age male riskloss muslim christian , vce(cluster session)
boottest treated, svmat 

 
reg Donation treated age education income10k finance male riskloss muslim christian distanceroad population, vce(bootstrap, cluster(session) reps(500) dots(1))
 est sto reg4
 
 reg Donation treated age education income10k finance male riskloss muslim christian distanceroad population, vce(cluster session)
boottest treated, svmat 

 
tobit Donation treated age education income10k finance male riskloss muslim christian distanceroad population, ll ul vce(bootstrap, cluster(session) reps(500) dots(1))
 est sto reg2
 
 tobit Donation treated age education income10k finance male riskloss muslim christian distanceroad population, vce(cluster session)
boottest treated, svmat 

 
 estout reg3 reg1 reg4 reg2  using "../Elaborations/tables/tableDONATION-revision-revision.tex", replace stardetach ///
cells("b(star fmt(3))" "se(par fmt(3))") substitute( _ "-" -cons "Constant") ///
indicate() nobase ///
style(tex) label collabels(,none) stats(N, labels("N.obs.")fmt(0)) ///
 prehead(\begin{tabular}{p{.45\textwidth}*{4}{r@{}l}} \hline \hline ///
"& \multicolumn{2}{c}{Model 1} & \multicolumn{2}{c}{Model 2}  \\") ///
posthead(\hline) prefoot(\hline) postfoot(\hline \multicolumn{9}{p{\textwidth}} ///
{\footnotesize{\textbf{Notes:} ///
Dependent variable: coins appropriated by the dictator. ///
Models 1 and 3: OLS regression, ///
models 2 and 4: left-censored Tobit regression. ///
All models include village fixed-effects. ///
Robust standard errors clustered at the village level. ///
Compared to models 1 and 2, models 3 and 4 controls for income, finance, and education levels. /// 
Controls include: age, ///
 gender, ///
religion, ///
estimation of risk preferences. /// 
Symbols $***$, $**$, and $*$ indicate significance at the 1\%, 5\% and 10\% level, respectively.}} \end{tabular}) ///
starlevels(* 0.10 ** 0.05 *** 0.01) ml(,none) eql(,none)
******************************************************************************************************









